Text-to-Speech

This example uses the text-to-speech node from the Sherpa extension. The generated speech audio is rendered in real-time on the device's speaker.

JSON

{
    "type": "RealTimeGraphRenderer",
    "config": {
        "graph": {
            "config": {
                "sampleRate": 16000,
                "bufferSize": 512
            },
            "nodes": [
                {
                    "id": "ttsNode",
                    "type": "Sherpa.SherpaTTSNode"
                },
                {
                    "id": "monoToMultiChannelNode",
                    "type": "MonoToMultiChannel"
                }
            ],
            "connections": [
                {
                    "sourceNode": "ttsNode",
                    "destinationNode": "monoToMultiChannelNode"
                },
                {
                    "sourceNode": "monoToMultiChannelNode",
                    "destinationNode": "outputNode"
                }
            ]
        }
    }
}

To generate speech with the TTS node, call the synthesize action and provide the desired text as a parameter.

C++

auto result = SwitchboardV3::callAction("ttsNode", "synthesize", { { "text", text } });